LSHTM Research Online

Hunt, Luke; Gupta-Wright, Ankur; Simms, Victoria; Tamba, Fayia; Knott, Victoria; Tamba, Kon-goneh; Heisenberg-Mansaray, Saidu; Tamba, Emmanuel; Sheriff, Alpha; Conteh, Sulaiman; +6 more...Smith, Tom; Tobin, Shelagh; Brooks, Tim; Houlihan, Catherine; Cummings, Rachael; Fletcher, Tom;(2015) Clinical presentation, biochemical, and haematological parameters and their association withoutcome in patients with Ebola virus disease: an observational cohort study. The Lancet infectiousdiseases, 15 (11). pp. 1292-1299. ISSN 1473-3099 DOI: https://doi.org/10.1016/S1473-3099(15)00144-9

Downloaded from: http://researchonline.lshtm.ac.uk/id/eprint/2274327/

DOI: https://doi.org/10.1016/S1473-3099(15)00144-9

Usage Guidelines:

Please refer to usage guidelines at https://researchonline.lshtm.ac.uk/policies.html or alternativelycontact researchonline@lshtm.ac.uk.

Available under license: Copyright the publishers

https://researchonline.lshtm.ac.uk

Clinical Presentation, Biochemical and Haematological Parameters and their Association with Outcome in 1

Patients with Ebola Virus Disease: An observational Cohort Study 2

3

Short title: 4

Laboratory abnormalities in EVD 5

Authors: 6

Luke Hunt (MBChB)1, Ankur Gupta-Wright (MSc)2, Victoria Simms (PhD)3, Fayia Tamba (HND*)4, Victoria 7

Knott (MBChB)1, Kongoneh Tamba (HND*)4, Saidu Heisenberg-Mansaray (HND*)4, Emmanuel Tamba 8

(HND*)4, Alpha Sheriff (Nil)1, Sulaiman Conteh (MBChB)1, Tom Smith (MSc)1, Shelagh Tobin (BSc)5, Tim 9

Brooks (MBChB)6, Catherine Houlihan (MSc)2, Rachael Cummings (MSc)1 and Tom Fletcher (MRCP)5, 7 10

*Higher National Diploma 11

Author affiliations: 12

1. Save the Children International, St. Vincent House, 30 Orange Street, London WC2H 7HH 13

2. Department of Clinical Research, London School of Hygiene & Tropical Medicine, Keppel Street, 14

London WC1E 7HT 15

3. Department of Infectious Disease Epidemiology, London School of Hygiene & Tropical Medicine, 16

Keppel Street, London WC1E 7HT 17

4. Ministry of Health and Sanitation, 4th Floor Youyi Building, Freetown, Sierra Leone 18

5. Defence Medical Services, Whittington Barracks, Lichfield, Staffordshire WS14 9PY 19

6. Public Health England, Porton Down, Salisbury, Wiltshire SP4 0JG 20

7. Liverpool School of Tropical Medicine, Pembroke Place, Liverpool, Merseyside L3 5QA 21

22

Correspondence: 23

Luke Hunt 24

Luke.hunt@doctors.org.uk 25

(0044) 7786 397984 26

Keywords: 27

Ebola, Zaire Ebolavirus, viral hemorrhagic fever, laboratory abnormalities, haematology, biochemistry, case 28

management, outcomes 29

Word count: Abstract: 249; Manuscript: 3000. 30

31

Abstract 32

Background 33

Clinical management of Ebola Virus Disease (EVD) remains challenging. Routine laboratory analytics are 34

often unavailable in the outbreak setting, and few data exist on the associated haematological and 35

biochemical abnormalities. We present laboratory and clinical data from the Kerry Town Ebola Treatment 36

Centre in Sierra Leone to better inform clinical management algorithms, improve understanding of key 37

variables associated with outcome and provide insight into the pathophysiology of EVD. 38

Methods 39

150 patients with confirmed EVD were admitted between 8th December 2014 and 9th January 2015. At 40

admission, all patients had clinical presentation recorded and blood taken for Ebola confirmation using 41

reverse-transcriptase–polymerase-chain-reaction (RT-PCR) and for haematological and biochemical 42

analysis. The association between these and clinical outcome was evaluated. 43

Findings 44

The mean age of cases was 26 years. Case fatality rate was 35% (55/150). Most patients presented with 45

stage 2 (gastrointestinal involvement, 61%, 72/118) and stage 3 (severe/complicated, 10%, 12/118) 46

disease. Acute Kidney injury (AKI) was common (50%, 52/104), as were abnormal serum potassium (33%, 47

32/97), severe hepatitis (59%, 54/92) and raised CRP (21%, 21/100). Haematological abnormalities were 48

common, including raised haematocrit (15%, 15/100), thrombocytopaenia (45%, 47/104) and 49

granulocytosis (42%, 44/104). Severe AKI, low RT-PCR cycle threshold (<20 cycles) and severe hepatitis 50

were independently associated with mortality. 51

Interpretation 52

EVD is associated with a high prevalence of haematological and biochemical abnormalities, even in mild 53

disease and in the absence of gastrointestinal symptoms. Clinical care targeting hypovolaemia, electrolyte 54

disturbance and AKI are likely to reduce historically high case fatality rates. 55

Funding 56

None 57

58

59

Introduction 60

Ebola virus disease (EVD) is caused by an RNA Filovirus, and is characterised by a febrile illness (87%), often 61

progressing to diarrheoa (67%), vomiting (66%) and, sometimes, haemorrhage (18%).1 The current 62

outbreak in West Africa is the largest the world has seen, with almost 25,000 cases (25th March 2015).2 63

Clinical and public health management is challenging, especially in the worst hit countries of Sierra Leone, 64

Liberia and Guinea, due to severity of clinical presentation, infection control concerns, poor health-system 65

infrastructure and high population density.3,4 66

67

There are currently no proven treatments for EVD, thus management is supportive, including 68

administration of oral and intravenous fluids and electrolytes, management of co-infections and symptom 69

control.3-5 EVD is recognised to cause marked biochemical abnormalities which may be amenable to simple 70

interventions, potentially reducing the high case fatality rate (CFR).4,6-12 Due to the low-resource setting of 71

outbreaks and risks associated with collecting and processing laboratory samples, few data exist on these 72

abnormalities, mostly from animal models, small cohorts or case studies from resource-rich settings.6-9,11,13 73

74

Kerry Town Ebola Treatment Centre (ETC), operated by Save the Children, opened on November 5th 2014 in 75

Sierra Leone offering supportive care focused on fluid and electrolyte management, in accordance with 76

World Health Organization (WHO) guidelines.3 This included haematology and biochemistry testing for all 77

admissions from December 8th 2014. We report data on clinical presentation, laboratory abnormalities, and 78

their association with mortality in patients admitted over a one month period. This data will assist in case 79

management in centres without access to laboratory support, provide insight into the disease 80

pathophysiology, corroborate animal models, and help identify priority areas for future research. 81

82

Methods 83

Study Design 84

We conducted a cohort study of patients consecutively admitted to Kerry Town ETC between 8th December 85

2014 and 9th January 2015. During this period only patients with EVD confirmed at other isolation or 86

community care centers were admitted. We used a standard case definition as per WHO guidelines.2 All 87

patients were included in the study except for those who were dead on arrival, or those who had no blood 88

results within 24 hours of admission. Primary outcome measure was discharge from the ETC. 89

90

Blood Analysis 91

Samples were analysed at the on-site laboratory run by Public Health England and the UK Defence Medical 92

Services (UK-DMS). The Piccolo Express® system (Abaxis, California, USA) was used to generate metabolic 93

and liver function profiles. An assay with biochemistry and liver function profile was primarily used 94

(Amylyte 13), though some samples were processed on a separate assay that included lactate, magnesium, 95

calcium and phosphate but excluded liver function profile (Metlac 12). Laboratory scientists selected the 96

assay based on availability, unless clinically requested. Haematology samples were processed using a 97

Horiba ABX Micros ES60® analyser (Horiba, Montpellier, France). Reverse-transcriptase-polymerase-chain-98

reaction (RT-PCR) assays for diagnosis of EBOV were performed using the Altona RealStar® Filovirus RT-PCR 99

Kit (Altona, Hamburg, Germany), following inactivation and manual RNA extraction. Positive results were 100

reported as Ct (cycle threshold) values. All patients had a repeat RT-PCR on admission. 101

102

Data Collection 103

All data was collected as part of routine patient care, and recorded on standardised forms, which were kept 104

securely. Data extracted for research purposes was anonymised and stored on a password-protected 105

database. The Sierra Leone Ethics and Scientific Review Committee provided study approval. 106

107

Clinical Staging and management 108

All patients were assigned a disease stage on admission according to their clinical features. The staging 109

system was developed by the UK-DMS in reference to WHO guidelines and the proposed pathogenesis of 110

EVD and is outlined in Table 1.3,12-15 Children were staged using a similar system, informally adapted to 111

include paediatic indicators of dehydration as per WHO.3 Standardised therapy was administered to all 112

patients based on disease stage. Targeted electrolyte replacement was provided based on biochemistry 113

results. Patients were discharged once asymptomatic for 72 hours with negative repeat blood Ebola RT-114

PCR. Acute Kidney Injury (AKI) was defined according to RIFLE criteria.16 To estimate baseline creatinine, we 115

used the Modification of Diet in Renal Disease (MDRD) for adults,17 and the Schwartz calculation for 116

children.18 For ALT and AST, ‘high’ was defined as 5-15 times the upper limit of normal (ULN), and ‘very 117

high/severe’ was defined as >15 times ULN. EBOV RT-PCR Ct value (Ct) was categorised as low (<20 cycles) 118

or high (≥20 cycles) with low Ct indicating high viral load. 119

120

Statistical Analysis 121

Descriptive analyses are reported as frequencies, proportions, means or medians. We used χ2, t-test and 122

Wilcoxon rank-sum for comparisons. Risk factors for mortality were assessed using logistic regression. A 123

priori variables, and variables associated with mortality in univariate analysis (at p<0.1) were assessed in a 124

multivariate model. Interactions were assessed using likelihood ratio tests. Adjustment was not made for 125

missing data since biochemical and haemotological variables were missing at random. Hypothesis tests 126

were two-tailed (at p<0·05). Data was analysed using Stata (Statacorp, Texas, USA), v13. 127

128

Role of the Funding Source 129

The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or 130

writing of the report. The corresponding author had full access to the data in the study and had final 131

responsibility for the decision to submit for publication. 132

133

Results 134

During the study period, 150 patients were admitted with confirmed EVD. Additionally 6 non-EVD cases 135

were referred and 4 patients died in transit. Absent laboratory samples were due to no sample being 136

obtained on admission (n=8) and a laboratory isolator fault meaning no samples were processed from 26th-137

29th December (n=24). Of the 118 included patients, 104 had haematology and 114 had biochemistry 138

analysed. In these patients, absent tests were due to no haematology sample being taken (n=12), 139

haematology sample inadequate for analysis (n=2) or no biochemistry sample taken (n=4). 101 patients had 140

liver function analysed and 18 had lactate measured, with 5 having both. 141

142

Baseline characteristics are described in Table 2. Among EVD cases, 52·5% were male. The mean age was 143

25·9 years (sd 14·7). Children (<18 years) accounted for 32·2%. Median time from admission to discharge 144

amongst survivors was 8 days (inter quartile range (IQR) 5-12) and median time from admission to death 145

was 4 days (IQR 3-5). Patients without laboratory investigations were more likely to be under 5 years 146

(p=0·003) and female (p=0·033) but otherwise had similar baseline characteristics to those with laboratory 147

results. 148

149

Mortality 150

Overall CFR for the study period was 36·7% (55/150). CFR in those included in this study was 34·7% 151

(41/118). CFR was non-significantly higher in men than women (40·3% vs. 28·6%. X2=1·79, p=0·181) and 152

lower in children (23·7% vs. 40·0%, X2=3·02, p=0·082). Higher disease stage on admission was strongly 153

associated with mortality (66·7%, 32·7% and 25·8% for stage 3, 2 and 1 respectively, p=0·001). Mean time 154

from symptom onset to admission was shorter in patients who died than in survivors. There was no 155

association between time from symptom onset and disease stage at admission. 156

157

RT-PCR Ct values 158

The mean Ct in positive patients was 21·4 cycles (sd 4·5), with 41·1% (46/112) having a low Ct. Low Ct on 159

admission was strongly associated with mortality (65·2% vs. 13·6%, p<0·001). Fifteen patients were 160

transferred following positive EBOV RT-PCR tests at the referring center, but tested negative at admission, 161

despite remaining symptomatic. Their survival rate was 100%. 162

163

Biochemistry 164

Biochemistry results for the cohort are outlined in Table 3 and Figure 1. Creatinine levels were significantly 165

higher in fatal than non-fatal cases (median 467µmol/L vs. 90µmol/L respectively, p<0·001, n=104). AKI was 166

common, occurring in 50·0% (52/104) of cases. AKI occurred more commonly in fatal than non-fatal cases 167

(81·3% vs. 18·1%, p<0·001) and when it did occur it was always severe (RIFLE-3). Incidence of AKI did not 168

differ significantly by clinical stage (χ2= 9·83, p=0·610). Notably, 28·6% (9/31) of patients admitted with 169

stage 1 disease had severe AKI. 170

171

Abnormal potassium occurred in 33·0% (32/97) of admissions, with similar proportions having 172

hypokalaemia and hyperkalaemia (19·6%, 19/97 and 13·4%, 13/97 respectively). Patients that died were 173

more likely to have hyperkalaemia (35·7% vs. 4·4% of survivors, p<0·001), as were patients with AKI (26·7% 174

with RIFLE-3 AKI vs. 6·5% with no AKI or RIFLE 1-2 AKI, p=0·007). Hyponatraemia was common (31·9%, 175

36/113), but was not associated with outcome. Abnormal liver function was common, with 70·3% (71/101) 176

having alanine transaminase (ALT) or aspartate transaminase (AST) >5 times ULN. Severe hepatitis (AST >15 177

ULN) was more common in fatal than non-fatal cases (92·9% vs. 43·8% respectively, p<0·001). Median 178

AST:ALT ratio was 2·3, and did not differ significantly between outcomes. Bilirubin was normal in 87·0% of 179

cases. 180

181

Raised creatinine kinase (CK) was common, and higher in fatal than non-fatal cases (median CK 2938 IU/L 182

versus (vs.) 1519 IU/L, p=0·019, n=100). All cases with a normal CK survived. Fourteen cases had CK>5000 183

IU/L (upper limit of our assay). Almost all cases with RIFLE-3 AKI had raised CK (97·1% vs. 78·3% with no AKI, 184

p=0·014). Median C-reactive protein (CRP) was higher in fatal than non-fatal cases (median CRP 12.9 vs. 185

69.5, p<0.001, n=100), and more fatal cases had CRP >100mg/L (46·9% vs. 8·8% in survivors, p<0·001). 186

Lactate was raised in 91·7% (16/17) of cases. 187

188

Haematology 189

Haematology results for the cohort are outlined in Table 3 and Figure 1. Higher mean haemoglobin (Hb), 190

haematocrit (Hct) and median platelet count were associated with mortality. Mean Hb, Hct and platelets in 191

non-survivors versus survivors were 15·5g/dl vs. 13·5g/dl (p<0.001, n=103) 45·1 vs. 39·4% (p<0.001, n=100) 192

and 146 x109/L vs. 197 x109/L (p=0·005, n=104) respectively. Thrombocytopaenia was more common 193

amongst non-fatal cases (53·7% vs. 29·7%, p=0·019). Severe thrombocytopaenia (<50x109/L) was 194

uncommon (2·9%, 3/104). 195

196

Low median white cell count (WCC), lymphocyte count and granulocyte count also predicted survival. 197

Median WCC was 15·8 x109/L vs. 7·9 x 109/L, (p<0·001, n=104) median lymphocytes were 4·2 x109/L vs. 2·2x 198

109/L (p<0·001, n=104) and granulocytes were 11·5 x 109/L vs. 4·3 x 109/L (p=0.001, n=104) in non-survivors 199

vs. survivors respectively. Granulocytosis was present in 42·3% (44/104) of cases, and lymphocytosis was 200

more common in fatal cases (66·7% vs. 31·0% p=0·002). Malaria co-infection was uncommon with 2·5%, 201

3/114 having a positive Malaria RDT on admission. 202

203

Risk factors for mortality 204

In unadjusted analyses, the strongest risk factors for mortality were RIFLE-3 AKI (OR 19·7, 95%CI 6·7-57·4, 205

vs. no AKI or RIFLE-1-2 AKI, p<0·001), severely raised AST (OR 16·7, 95%CI 3·7-76·5, vs. not severely raised 206

AST, p<0·001), high haematocrit (OR 13·4, 95%CI 2·7-66·7, vs. normal, p=0·002), low EBOV RT-PCR Ct (OR 207

11·9, 95%CI 4·7-30·1, vs. high Ct, p<0·001), hyperkalaemia (OR 11·1, 95%CI 2·7-45·7, vs. normal potassium, 208

p=0·001), CRP >100mg/dl (OR 9·1, 95%CI 3·1-27·1, vs. CRP ≤100, p<0·001) and granulocytosis (OR 8·7, 209

95%CI 2·9-26·5, vs. normal granulocytes, p<0·001). 210

211

Although hyperkalaemia appeared to be associated with RIFLE-3 AKI, both remained associated with 212

mortality in bivariate analyses (p<0·001). Similarly, haemoglobin and haematocrit were associated with 213

each other, however haematocrit remained strongly associated with mortality after adjusting for 214

haemoglobin (OR 26·9, 95%CI 2·5-289·3). CRP remained strongly associated with mortality (OR 16·5, 95%CI 215

3·8-71·2), even after adjustment for granulocytosis. 216

217

Multivariate analysis was undertaken using variables associated with mortality a priori (age, gender, RT-PCR 218

Ct) or in univariate/bivariate analysis. Age was fitted as a continuous variable. Hyperkalaemia, CRP and 219

granulocytosis appeared to become protective due to collinearity and were removed to improve stability. 220

Hematocrit was removed because it had no effect. No interaction terms contributed to the model. The final 221

model is shown in Table 4. The strongest independent risk factors for mortality were low EBOV RT-PCR Ct 222

(OR 6·7, p=0·013, 95% CI 1·5-30.1) and RIFLE-3 AKI (OR 5·8, p=0·033, 95% CI 1·2-29·6). Disease stage on 223

admission was not associated with mortality after adjustment for other risk factors. 224

225

Discussion 226

The natural history of EVD is well characterised with a gastrointestinal stage leading to significant 227

hypovolaemia, systemic hypoperfusion and shock in inadequately fluid-resuscitated patients.4 The extent of 228

multi-organ dysfunction syndrome—commonly involving the renal, hepatic, neurological and coagulation 229

systems—varies from mild dysfunction to irreversible organ failure and death, but the pathogenesis of EVD 230

remains poorly understood. Key to improving outcomes is the availability of routine biochemistry and 231

haematological testing. Large observational studies are also required to compare data with animal models 232

and support clinical trials of novel therapeutics. 233

234

The majority of our patients were admitted with stage 2 or 3 disease (74/118), and had high levels of organ 235

dysfunction and associated metabolic abnormalities. Fifty percent (52/104) of cases had AKI, and raised 236

creatinine was associated with mortality, supporting earlier published data.6, 7 Importantly kidney 237

dysfunction was not limited to later disease stage, supporting the role of IV fluid therapy in early disease. 238

239

AKI is generally attributed to pre-renal causes.4,15 However, our results suggest AKI may be multifactorial, 240

given the high prevalence in early disease where dehydration is a less common clinical finding. The role of 241

rhabdomyolysis is unclear and worthy of further study given our almost universal findings of elevated CK 242

(and its association with AKI), and disproportionately elevated AST. However CK was rarely elevated to 243

levels associated with AKI, apart from a discreet number of patients who reached the ceiling of our assay 244

measurement. Subsequent renal injury may also arise due to poor renal perfusion secondary to profound 245

hypovolaemia from gastro-intestinal losses, or viral or secondary bacterial sepsis resulting in acute tubular 246

necrosis.19 Microvascular damage from disseminated intravascular coagulation may also contribute; raised 247

D-Dimer was a common finding in patients with Sudan Ebolavirus6 and autopsy of infected primates 248

reported fibrinous cellular debris in glomerulus.20 Importantly, our data show that EVD-associated AKI can 249

be managed successfully in this setting. 250

251

Hepatocellular inflammation has previously been reported.8,9 Our data showed that most patients had 252

elevated ALT/AST levels, which was associated with mortality. Normal bilirubin levels indicated fulminant 253

hepatic failure is not a major feature consistent with the clinical syndrome observed. Elevated CRP and 254

granulocytosis were also associated with mortality, which may reflect secondary bacterial infection, 20 255

although all patients were treated with a third-generation cephalosporin from stage 2 of clinical disease. 256

257

The data is the first reported on haematological abnormalities, and show that mortality was associated 258

with higher haemoglobin and haematocrit. This may be a surrogate marker for intravascular fluid depletion, 259

widely believed to be associated with poor outcomes.4 Thrombocytopaenia was common and occurred in 260

45·2% of patients at presentation, but was rarely severe. It was not possible to measure coagulation in our 261

cohort, and this should be a priority for future research given the frequency and poorly understood nature 262

of haemorrhage in EVD. Semi-quantitative EBOV viral load measurement has been shown to be a useful 263

predictor of mortality 8,12 and was associated with clinical outcome in our series. 264

265

We report a number of study limitations. The CFR is lower than others have reported. 7,10,11 While this may 266

be due to a high-standard clinical care, it is possible that there is a survivor bias (i.e. patients with EVD more 267

likely to survive were admitted to this ETC) since only patients confirmed to have Ebola were admitted 268

during the study period, and there were time delays prior to ETC admission. This is supported by the finding 269

that some patients were RT-PCR negative at admission, suggesting that they may have already cleared the 270

virus. Caution should be used in comparing CFRs between settings. The finding that a shorter time from 271

symptom onset to admission was associated with mortality may have been caused by recall bias and 272

survivor bias; symptom onset date was collected from records sent with the patient, which were not always 273

complete. Patients also presented to the ETC at different stages of illness, and had variable levels of 274

treatment in holding centres prior to admission. 275

276

Due to an isolator failure, 24 consecutive patients had no laboratory testing. The significance of this to our 277

results is unclear. A number of the assays (AST, ALT, CK) were limited by their upper limits of detection, 278

potentially underestimating the difference between groups. Additionally, baseline creatinine was estimated 279

based on MDRD scores, which may have over- or under-estimated the prevalence of AKI. 280

281

Conclusion 282

Our data show that the provision of routine laboratory support to an ETC provides opportunity for 283

improved understanding of disease pathophysiology, and correlation with clinical disease staging. We 284

demonstrate a previously unreported high prevalence of electrolyte imbalance and organ dysfunction in all 285

stages of EVD. The ability to diagnose and treat these abnormalities at presentation and follow patients’ 286

response to therapy is likely to improve survival. 287

288

The most important aspects of supportive care are aggressive management of intravascular volume 289

depletion, correcting electrolyte abnormalities, and preventing the complications of shock.21 These are 290

underlying tenets of critical care medicine, which can and should be applied in this setting. Improving the 291

provision of good supportive care including laboratory analysis is essential to further clarify the 292

pathogenesis and pathophysiology of EVD in humans. In parallel to ongoing intervention studies, this will 293

inform evidence-based protocols to improve outcomes for this outbreak and the next. 294

295

Acknowledgements 296

We would like to thank Save the Children for enabling us to collect the data, and to all staff at Kerry Town 297

ETC for their contribution towards the successful running of the facility. 298

299

Disclosures 300

None 301

302

Author Contribution Statement 303

LH, AGW, VS, FT, VK, CH, RC and TF contributed towards study design. LH, AGW, VS, FT, VK, KT, SHM, ET, 304

AS, SC, TS, ST, TB and TF contributed towards data collection. AGW, VS, ST and TB contributed towards 305

data analysis. LH, AGW, VS and TF drafted the manuscript. All authors contributed towards manuscript 306

revision. 307

308

309

310

Table 1 311

Stage Clinical Features Typical Patient Standard Therapy

1 Early/Mild Non-specific features. Pyrexia, weakness, lethargy, myalgia, arthritis

Ambulatory, able to compensate for fluid losses via oral intake

Oral Rehydration Solution Symptomatic treatment Zinc / Multivitamins Anti-malarials if RDT positive Targeted electrolyte replacement+

Treatment of hypoglycemia

2 Gastrointestinal Involvement

As above plus: diarrhoea, vomiting and/or abdominal pain

Unable to compensate for fluid losses via oral intake due to emesis or loss of large volumes

As per stage 1, plus: IV fluid therapy (3000-6000ml/24h for adults, guided by fluid and electrolyte balance) IV Ceftriaxone*

3 Complicated As above plus: hemorrhage, shock, neurological involvement and/or signs of organ failure

Critically ill, usually hypovolaemic, often with confusion and/or seizures, bleeding.

As per stage 1+2, plus: As clinically indicated: Sedation/anti-epileptics Vitamin K Fresh Frozen Plasma#

RDT rapid malaria diagnostic test; IV intravenous. + Sodium, potassium, magnesium, calcium and phosphate replaced according to 312 biochemistry results. *Ceftriaxone (2g once per day for adults) was given as an empirical 5 day course in all stage 2 and 3 patients. 313 It was continued for longer or given to stage 1 patients if clinically indicated. # Fresh frozen plasma was given when hemorrhage 314 occurred and when available. 315 316

Table 1 Clinical staging system for Ebola virus disease used at Kerry Town ETC and subsequent standard 317

clinical management. 318

Table 2 319

Included in study No laboratory tests sent

Survived Died Total

Total number

77 41 118 32

Gender Male 37 (48·1) 25 (61·0) 62 (52·5) 10

Female 40 (51·9) 16 (39·0) 56 (47·5) 22

Age (years) Mean (sd) 23·2 (13·1) 31·0 (16·2) 25·9 (14·7) 25·1 (18·5)

0-4 3 (3·9) 0 3 (2·5) 5

5-14 20 (26·0) 8 (19·5) 28 (23·7) 5

15-24 20 (26·0) 5 (12·2) 25 (21·2) 4

25-34 20 (26·0) 11 (26·8) 31 (26·3) 9

35-44 7 (9·1) 8 (19·5) 15 (12·7) 5

45-80 7 (9·1) 9 (22·0) 16 (13·6) 4

Time from symptom onset to admission (days)

Mean (sd) 5·3 (3·2) 4·4 (2·9) 5·0 (3·1) 5·6 (2·2)

1-3 16 (20·8) 13 (31·7) 29 (24·6) 4

4-6 23 (29·9) 14 (34·1) 37 (31·4) 12

≥7 14 (18·2) 3 (7·3) 17 (14·4) 7

Unknown 24 (31·2) 11 (26·8) 35 (29·6) 9

Stage at admission

Stage 1 23 (29·9) 8 (19·5) 31 (26·3) 7

Stage 2 47 (61·0) 25 (61·0) 72 (61·0) 16

Stage 3 4 (5·2) 8 (19·5) 12 (10·2) 7

Unknown 3 (3·9) 0 3 (2·5) 2

Ebola RT-PCR at admission

Positive 62 (80·5) 41 (100·0) 103 (87·3) 30

Negative 15 (19·5) 0 15 (12·7) 1

No result 0 0 0 1

Malaria RDT Positive 1 (1·3) 2 (4·9) 3 (2·5) 1

Negative 72 (93·5) 39 (95·1) 111 (94·1) 29

Unknown 4 (5·2) 0 4 (3·4) 2 Data are numbers (column %) unless otherwise stated. RT-PCR reverse-transcriptase-polymerase-chain-reaction; RDT rapid malaria 320 diagnostic test. 321 322

Table 2 Characteristics of patients with confirmed Ebola virus disease admitted to Kerry Town ETC during 323

the study period 324

Table 3 325

Haematology

Survived Died Total Test

Haemoglobin Mean g/dL (sd) 13·5 (2·1) 15·5 (2·6) 14·2 (2·5) p<0·001*

Anemia n (%) 20/66 (30·3)

5/37 (13·5)

25/103 (24·3)

p=0·057#

Haematocrit

Mean % (sd) 39·4 (6·6) 45·1 (7·5) 41·4 (7·5) p<0·001*

Raised, n(%) 2/64 (3·1) 13/36 (36·1)

15/100 (15·0)

p<0·001#

Low, n(%) 29/64 (45·3)

7/36 (19·4)

36/100 (36·0)

Platelets

Median x109/L (IQR) 146 (108-219)

197 (145-346)

155 (119-247)

P=0·005$

Thrombocytopaenia, n(%)

36/67 (53·7)

11/37 (29·7)

47/104 (45·2)

p=0·019#

White blood cells

Median x109/L (IQR) 7·9 (4·8-12·4)

15·8 (8·6-21·7)

10·0 (5·2-16·0)

p<0·001$

Raised, n(%) 21/67 (31·3)

25/37 (67·6)

46/104 (44·2)

p=0·001#

Low, n(%) 12/67 (17·7)

5/37 (13·5)

17/104 (16·4)

Lymphocytes

Median x109/L (IQR) 2·2 (1·5-3·7)

4·2 (2·2-6·6)

2·6 (1·6-4·4)

p<0·001$

Raised, n(%) 19/67 (28·4)

24/37 (64·9)

43/104 (41·4) p=0·001#

Low, n(%) 7/67 (10·4) 1/37 (2·7) 8/104 (7·7)

Granulocytes

Median x109/L (IQR) 4·3 (2·5-8·0)

11·5 (4·9-14·0)

5·7 (2·8-11·0)

p=0·001$

Raised, n(%) 19/67 (28·4)

25/37 (67·6)

44/104 (42·3)

p<0·001#

Low, n(%) 15/67 (22·4)

7/37 (18·9)

22/104 (21·2)

Viral Load

RT-PCR Cycle time

Mean n (sd)*

23·4 (4·3), 60

18·3 (2·7), 39

21·4 (4·5), 99

p<0·001*

Low, n(%) 16/73 (21·9)

30/39 (76·9)

46/112 (41·1)

p<0·001#

Biochemistry

Creatinine

Median µmol/L (IQR)

90 (58-161)

467 (225-754)

121 (65-392) P<0·001$

RIFLE-1, n(%) 5/72 (6·9)

0/32 5/104 (4·8)

p<0·001# RIFLE-2, n(%) 8/72 (11·1)

0/32 8/104 (7·7)

RIFLE 3, n(%) 13/72 (18·1)

26/32 (81·3)

39/104 (37·5)

Potassium

Mean mmol/L (sd) 4·0 (0·7) 4·8 (0·9) 4·2 (0·9) p<0·001*

Raised, n(%) 3/69 (4·4)

10/28 (35·7)

13/97 (13·4) p<0·001#

Low, n(%) 16/69 (23·2)

3/28 (10·7) 19/97 (19·6)

Sodium

Mean mmol/L (sd) 129·8 (4·6)

128·5 (5·8) 129·4 (5·1) p=0·182*

Low, n(%) 22/76 (30·0)

14/37 (37·8)

36/113 (31·9)

p=0·341#

Lactate Mean mmol/L (sd) 4·0 (1·4) 4·5 (0·6) 4·2 (1·2) p=0·521*

Raised, n(%) 11/12 (91·7)

5/5 (100·0) 16/17 (94·1) p=0·506#

Total dissolved CO2

Median mmol/L (IQR)

22 (17-28)

17 (11-18) 20 (15-23) P=0·035$

Low, n(%) 3/12 (25·0)

4/6 (66·7) 7/18 (38·9) p=0·087#

Liver function tests

Alanine transaminase (ALT)

Median IU/L (IQR) 192 (86-371)

577 (314-1195)

253 (121-594)

P<0·001$

High, n(%) 21/68 (30·9)

16/33 (48·5)

37/101 (36·6)

p<0·001#

Very high, n(%) 5/68 (7·4)

11/33 (33·3)

16/101 (15·8)

Aspartate transaminase (AST)

Median IU/L (IQR) 348 (129-1254)

>2000 (1491->2000)

867 (168->2000

P<0·001$

High, n(%) 11/64 (17·2)

2/28 (7·1) 13/92 (14·1) p<0·001#

Very high, n(%) 28/64 (43·8)

26/28 (92·9)

54/92 (58·7)

AST:ALT ratio Median (IQR), n 2·2 (1·4-3·3)

3·1 (1·6-3·5)

2·4 (1·4-3·4) p=0·259$

Creatinine kinase

Median IU/L (IQR), n 1519 (367-3454)

2938 (1137-3400)

1949 (589-3400)

P=0·019$

High, n(%) 51/68 (75·0)

32/32 (100·0)

83/100 (83·0)

p=0·002#

C-reactive protein

Median mg/L (IQR) 12·9 (5·0-29·4)

69·5 (22·5-147·0)

21·4 (6·0-77·0)

P<0·001$

High, n(%) 6/68 (8·8)

15/32 (46·9)

21/100 (21·0)

p<0·001#

All data are numbers (column %) unless otherwise stated· *median only presented for RT-PCR results that were positive·* denotes p-value from t-test, #deontes p-value from χ2 test, $deontes p-value from Wilcoxon rank-sum· Definitions are as follows; RT-PCR - reverse-transcriptase-polymerase-chain-reaction, anemia - as per WHO definitions, 22 thrombocytopaenia <150 x x109/L, raised white cell count >11 x109/L, low white cell count <4 x109/L, raised lymphocytes >3·2 x109/L, low lymphocytes <1 x109/L, raised granulocytes >7·5 x109/L, low granulocytes <2·5 x109/L, low RT-PCR Cycle time <20 cycles, RIFLE-1 AKI 1·5-2 x baseline creatinine, RIFLE 2 AKI 2-3 x baseline creatinine, RIFLE-3 AKI >3 x baseline creatinine, raised potassium >5·1 mmol/L, low potassium <3·6mmol/l, raised Lactate >2·1 mmol/l, low total dissolved CO2 <18 IU/L, high ALT >240 - ≤720 IU/L, very high ALT >720 IU/L, high AST >175 - ≤525 IU/L, very high AST >525 IU/L, high creatinine kinase >380 U/L and high c-reactive protein >100 mg/l·

Table 3 Laboratory results and abnormalities in patients admitted to Kerry Town ETC with confirmed Ebola

326

Figure 1 327

328

ALT alanine transaminase; AST asparate transaminase; CK creatinine kinase; CRP C-reactive protein; Cycle time is Ebolavirus 329 reverse-transcriptase-polymerase-chain reaction cycle time. P-values are from t-tests (potassium and sodium) and Wilcoxon rank-330 sum tests (all others). Shaded area indicates normal range for the UK based population. 331

Figure 1 Box plots of laboratory results from patients admitted to Kerry Town ETC; results are presented by 332

outcome. 333

334

335

Table 4 336

OR odds ratio; RT-PCR reverse-transcriptase-polymerase-chain-reaction; AST aspartate transaminase, AKI acute kidney injury. P-337 values are from likelihood ratio tests. Age was treated as a continuous variable. 338 339

Table 4 Factors associated with mortality in patients with previously confirmed Ebola virus admitted to 340

Kerry Town ETC 341

342

343

Univariate analysis Multivariate analysis

OR (95% CI) p-value OR (95% CI) p-value

Female Gender 0·59 (0·27-1·28) 0·182 0·33 (0·07-1·63) 0·174

Age 1·04 (1·01-1·07) 0·008 1·01 (0·95-1·06) 0·819

Disease stage 2.16 (1.08-4.32) 0.029 1.40 (0.41-4.84) 0.593

RT-PCR Cycle threshold <20

11·88 (4·69-30·06) <0·001 6·72 (1·50-30.07) 0·013

RIFLE-3 AKI 19·67 (6·73-57·43) <0·001 5·84 (1·15-29·58) 0·033

AST>525 U/L 16·71 (3·65-76·47) <0·001 6.95 (0·70-68·86) 0·097

Potassium >5·1 mmol/L

11·11 (2·70-45·66) <0·001 -

CRP >100mg/dl 9·12 (3·07-27·07) <0·001 -

Granulocytes >7·5 x109/L

8·68 (2·85-26·45) <0·001 -

High haematocrit 13·41 (2·70-66·67) 0·002 -

References:

1. WHO Ebola Response Team. Ebola virus disease in West Africa—the first 9 months of the epidemic and forward projections. New England Journal of Medicine. 2014. 371(16): 1481-1495.

2. World Health Organization. WHO Ebola Situational Report- 25th March 2015. Geneva: 2015.

3. World Health Organization. Clinical Management of Patients with Viral Haemorrhagic Fever: A Pocket Guide for the Front-Line Health Workers. Geneva: 2014.

4. Fowler R, Fletcher T, Fischer W, Lamontagne F, Jacob S, Brett-Major D et al. Caring for Critically Ill Patients with Ebola Virus Disease. Perspectives from West Africa. American Journal of Respiratory Critical Care Medicine. 2014; 190(7): 733-737.

5. Yazdanpanah Y, Arribas J, Malvy D. Treatment of Ebola Virus Disease. Intensive Care Medicine. 2014; 41(1): 115-117.

6. Rollin P, Bausch D, Sanchez A. Blood Chemistry Measurements and d ‐Dimer Levels Associated with Fatal and Nonfatal Outcomes in Humans Infected with Sudan Ebola Virus. Journal of Infectious Diseases. 2007; 196(s2): S364-S371.

7. Schieffelin J, Shaffer J, Goba A, Gbakie M, Gire S, Colubri A et al. Clinical Illness and Outcomes in Patients with Ebola in Sierra Leone. New England Journal of Medicine. 2014; 371(22): 2092-2100.

8. Sanchez A, Lukwiya M, Bausch D, Mahanty S, Sanchez A, Wagoner K et al. Analysis of Human Peripheral Blood Samples from Fatal and Nonfatal Cases of Ebola (Sudan) Hemorrhagic Fever: Cellular Responses, Virus Load, and Nitric Oxide Levels. Journal of Virology. 2004; 78(19): 10370-10377.

9. Kortepeter M, Bausch D, Bray M. Basic Clinical and Laboratory Features of Filoviral Hemorrhagic Fever. Journal of Infectious Diseases. 2011; 204(s3): S810-S816.

10. Bah E, Lamah M, Fletcher T, Jacob S, Brett-Major D, Sall A et al. Clinical Presentation of Patients with Ebola Virus Disease in Conakry, Guinea. New England Journal of Medicine. 2015; 372(1): 40-47.

11. Chertow D, Kleine C, Edwards J, Scaini R, Giuliani R, Sprecher A. Ebola Virus Disease in West Africa — Clinical Manifestations and Management. New England Journal of Medicine. 2014; 371(22): 2054-2057.

12. Rees P, Lamb L, Nicholson-Roberts T, Ardley C, Bailey M, Hinsley D et al. Safety and Feasibility of a Strategy of Early Central Venous Catheter Insertion in a Deployed UK Military Ebola Virus Disease Treatment Unit. Intensive Care Medicine. 2015; 1-9

13. Davis K, Anderson A, Geisbert T, Steele K, Geisbert J, Vogel P. Pathology of Experimental Ebola Virus Infection in African Green Monkeys: Involvement of Fibroblastic Reticular Cells. Archives of Pathology & Laboratory Medicine. 1997; 121: 805-19.

14. Zaki S, Goldsmith C. Pathologic Features of Filovirus Infections in Humans. Current Topics in Microbiology and Immunology 1999; 235: 97-116.

15. Fletcher T, Fowler R, Beeching N. Understanding Organ Dysfunction in Ebola Virus Disease. Intensive Care Medicine. 2014; 40(12): 1936-1939.

16. Hoste E, Clermont G, Kersten A, Venkataraman R, Angus D, De Bacquer D, Kellum J. RIFLE Criteria for Acute Kidney Injury are Associated with Hospital Mortality in Critically Ill Patients: a Cohort Analysis." Critical Care. 2006; 10.3: R73.

17. Levey A, Bosch J, Lewis J, Greene T, Rogers N, Roth D. A more Accurate Method to Estimate Glomerular Filtration Rate from Serum Creatinine: A New Prediction Equation. Annals of Internal Medicine. 1999; 130.6: 461-470.

18. Schwartz G, Haycock G, Edelmann C, Spitzer A. A Simple Estimate of Glomerular Filtration Rate in Children Derived from Body Length and Plasma Creatinine. Pediatics. 1976; 58.2: 259-263.

19. Schrier R, Wang W. Acute Renal Failure and Sepsis. New England Journal of Medicine. 2004; 351(2): 159-169.

20. Kreuels B, Wichmann D, Emmerich P, Schmidt-Chanasit J, de Heer G, Kluge S et al. A Case of Severe Ebola Virus Infection Complicated by Gram-Negative Septicemia. New England Journal of Medicine. 2014; 371(25): 2394-2401.

21. Rivers E, Nguyen B, Havstad S, Ressler J, Muzzin A, Knoblich B et al. Early Goal-Directed Therapy in the Treatment of Severe Sepsis and Septic Shock. New England Journal of Medicine. 2001; 345(19): 1368-1377.

22. World Health Organisation. Iron deficiency Anaemia : Assessment, Prevention and Control : A Guide for Programme Managers. Geneva: World Health Organisation; 2001.